https://act101.ai/blog Build-in-public technical deep-dives from the act101 team. Tue, 21 Jul 2026 00:00:00 +0000 https://act101.ai/blog/deleted-the-enum https://act101.ai/blog/deleted-the-enum An enum of nearly 180 variants and its 1,689-line dispatch match were textbook-correct human design, and hostile to agents. So I pointed the tool at its own source, it found its own worst structure, and I deleted it — the match down to 83 lines, then the 83 too. Tue, 21 Jul 2026 00:00:00 +0000 https://act101.ai/blog/cleanup-crew https://act101.ai/blog/cleanup-crew Agents made codebases decay faster than any human team ever could. The durable opportunity isn't generating more code — it's the cleanup layer that makes AI-generated code trustworthy to change. Tue, 14 Jul 2026 00:00:00 +0000 https://act101.ai/blog/agents-flying-blind https://act101.ai/blog/agents-flying-blind An agent's reasoning about your codebase can be flawless while its edits corrupt files, because it manipulates flat text instead of seeing the structure underneath. The fix is sight, not more intelligence. Tue, 07 Jul 2026 00:00:00 +0000 https://act101.ai/blog/inaugural https://act101.ai/blog/inaugural The origin story: an 8-hour, 50-million-token refactor broke me, so holed up in the desert I built the verification layer that had to exist before I'd let a single agent-written line ship. Wed, 01 Jul 2026 00:00:00 +0000